Data reported by Derbinski et al.1 reinforce the concept that thymic cells express a variety of tissue-restricted antigens and may be actively involved in negative selection. Based mainly on results obtained by flow cytometric analysis of cells derived from primary thymic tissue, the authors claim that the expression of a number of tissue-restricted proteins in medullary thymic epithelial cells (mTECS) appeared heterogeneous and that high expression of these molecules was restricted to a minor fraction of mTECS. One conclusion is that mTECs seem sufficient, but not necessary, to induce CD4+ T cell tolerance1.

In our view—and despite their conclusions—Derbinski et al. did not entirely rule out the possibility that other thymic cells, in addition to mTECS, can express a diverse range of tissue-specific antigens. Indeed, published observations show that some tissue-restricted antigens associated with autoimmune diabetes, including insulin and glutamic acid decarboxylase (GAD), are also expressed in the thymus by DCs and macrophages2,3. In addition, studies in mice have shown that the thymus contains cells, termed peripheral antigen–expressing (PAE) cells, that express insulin, and experiments in transgenic mice provide evidence that these PAE cells may have a tolerogenic function3,4,5,6. Similar cells expressing islet cell antigens such as insulin (proinsulin), GAD and the neuroendocrine antigen IA-2 have been described both in human thymic and peripheral lymphoid tissues3.

We favor the concept that microenvironmental effects—such as dynamic cytokine, growth factor and cell-cell contact fluxes in the thymus—are essential in modulating gene expression. Consequently, the complex cytokine and cell adhesion networks in the thymic microenvironment can modify the expression of tissue-restricted antigen expression. Similar effects have been shown in a number of other microenvironments7,8,9. In the study by Derbinsky et al. it is not clear whether DCs and other non-mTECs were cultured before sorting or whether tissue-restricted antigen expression by all cell populations under a variety of cytokine conditions that simulated different thymic microenvironments were evaluated1. We have compelling evidence that expression of tissue-restricted antigens—such as insulin and the ICA69 (islet cell autoantigen 69 kD10) in murine DCs—as well as insulin promoter activity in murine DCs varies dramatically depending on the cytokine mixture used in culture11; it can range from high expression to a complete absence (N. Giannoukakis et al. and M. Pietropaolo et al., unpublished data).

The thymic medulla provides a framework from which thymocytes can mature. Based on our findings, it would be reasonable to expect that not only mTECs, but also DCs, may play some role in either further nurturing or directing T cells to the deletion pathway within the thymic medullar compartment12. Derbinski et al. have provided some insights into this process, but more work is needed to determine whether different players within the thymic microenvironment influence the freedom of expression of their constituents.